Improving Drug Safety in Emergency Patients -a Randomized Controlled Trial

Overview

Aim/Objective: Investigate the effect of implementing a working model for performing medication reconciliation (MR) and medication review (MRe) in the emergency department (ED), on readmissions, patient safety and efficiency of the stay in the ED and the hospital. Research design: randomized, controlled, non-blinded trial. Control group; standard care. Intervention group; MR and MRe performed at admission to the ED by a clinical pharmacist in the interdisciplinary team. The intervention is based on a working model for MR, developed in our initiation project, and it will be adapted to also include MRe. Key challenges in this research field: Currently no implemented systematic model ensuring that the patient's correct medication list is obtained and assessed at the point of admission. There is lack of studies investigating the clinically outcome of performing MR and MRe in the ED. Lack of knowledge on the extent of drug related hospital admissions in Norway. These challenges are also recognized and prioritized by the Norwegian authorities. Impact and utility: The results from this study will give important answers to the challenges listed above. The results could imply a huge impact on how to organize ED in Norway regarding drug safety. If the hypothesis of this study is confirmed, implementing the intervention described will increase patient safety, both the hospital and society can reduce health care expenses from readmissions, and also the readmission-burden can be reduced for the patients.

Inclusion and randomization procedures Staff at the emergency department, including physicians and nurses will be informed about the project. At admission, the project pharmacist will describe the project to each potential participant and/or their next of kin, then provide written information about the project and answer potential questions. If patients temporary are unable to consent when asked to participate (e.g. delirium) their next of kin will be asked to supply a preliminary consent in the patients place. If the patient later refuses to participate he/she will be excluded from the trial, and any registered data for this patient will be deleted. Patients will periodically be included at day shift and evening shift and by different clinical pharmacists to reduce potential bias. We will randomize the patients into two study groups. The randomization process will be conducted by Department of Biostatistics and Epidemiology at Oslo University Hospital. This department will deliver randomization lists, and the project pharmacist will follow randomization procedure. Data registration Patient data will be registered on paper forms, which will be de-identified after the patient data is transferred de-identified to the password protected project database on the hospital research server. Only a code list will connect the patient to his or her data. Paper forms will at all times be kept accessible only to authorized project personnel, and eventually the forms will be maculated. De-identified patient information will not be brought out of the hospital. The code list connecting the patients to their data will at the latest be deleted 3 years after start of data collection. When results are published it will not be possible to identify individual patients. Customized Standard Operating Procedures addressing inclusion and randomization operations, registry operations and how to perform the intervention is developed. Sample size calculation Available information about readmission frequency at Diakonhjemmet Hospital is based on 30 days follow-up, and therefore cannot be used to calculate proportion of patients readmitted after 12 months. However, numbers from Oslo University Hospital estimate a readmission proportion of 50% after 12 months in a comparable patient population. Therefore this estimate is used as the expected readmission rate in the control group of this project. In a previous Swedish study conducted by Ulrika Gillespie who is member of the reference group of this project, a 16% reduction in hospital revisits within 12 months was found amongst older patients (\>80 years) following a comparable intervention as described in this project. On this basis, it will be necessary to include at least 146 patients in each group to show a significant effect on the primary endpoint (significance level of 5%, study power of 80%). However, the elderly patients included in the Swedish study had more comorbidity and therefore more use of health care resources. In this project we will include all patients 18 years and older and thereby the difference between the control group and intervention group probably will be smaller. A more realistic difference between the groups is 10%; thereby 385 patients would have to be included in each group to show a significant effect on the primary endpoint. To compensate for dropout the aim is to include 400 patients in each project group, thus a total of 800 patients. Based on statistics from Diakonhjemmet Hospital, inclusion of this amount of patients from the Emergency Department would require an inclusion period of 12 months. Statistics and analysis Statistical analyses will be conducted in IBM (International Business Machines)SPSS Statistics (Statistical Package for the Social Sciences). Data will be assessed for normality and analyzed according to appropriate statistical tests. The baseline demographic and clinical characteristics will be summarized using proportions, means and standard deviations, or median and interquartile range, as appropriate. Baseline comparisons: Characteristics of project participants in the two project groups will be compared using the chi-square test for categorical variables and the Student's t-test or non-parametric equivalent (e.g. the Mann-Whitney U test) for continuous variables. Multivariable analysis (logistic regression) will be used to compare endpoints between project groups while adjusting for prognostic variables and potential confounders. All statistical tests will be interpreted with a significance level of 5% (two-tailed). For building the model for prioritizing patients at increased risk of drug-related admissions and drug-related problems at admission to the emergency department binary regression analysis will be used. Data will be analyzed according to intention-to-treat (ITT) principles. In addition to ITT analysis, per protocol analysis will also be performed. Approval The project is approved by the Regional committee for medical and health research ethics (REC) and the research committee at Diakonhjemmet Hospital.